DS3-Datathon 2025 - 👑LeCoders

This repo is our submission to the anual 2025 DS3 Datathon.

With our Group name inspired by the basketball legend Lebron James!

This competion was from Feb 17-22, and consisted of 3 Problems

Our team placed 9th place, out of 180 participants, with a max group size of 4 members.

Warning

You will have to clone problem 2 and 3's datasets yoruself and place them inside their respective folders, as the datasets are to large to upload to github.

Problem A: Car Evaluation (Easy)

Group Accuracy: 95%

This problem, the eaiset of the 3 problems, involves getting a few data points about a vehnicle and predicting the cars saftey rating from that.

Kaggle Dataset

Problem B: Car Accident Analyasis (Med)

Group Accuracy: 93%

This problem is similar to the first question but you get accident data with various of different parameters ~30. And you have to predict which source 1 or 0, the data comes from. The data is requires cleaning and relevancy checks, we first disregard any entries with missing values, as the dataset was large enough. We only compute the source off the first 7 parameters in the dataset, this gave us 93% accuracy on our first test.

Kaggle Dataset

Problem C: Fungi Classification (Hard)

Group Accuracy: 22%

This problem consits of 5 different fungi types, trained on 1000 scrambled photos of each fungi type, the testing data is clean however which gave us some problems. This was the hardest of the 3 questions, we attempted to classify the images based on the frequency of pixel values and the hue but fell short due to time constraints.

Kaggle Dataset

This Datathon was our very first hands on experince in the field of Machine Learning and Data Science. We hope in the future with we are able to tackle the Fungi Problem with ease and have better success.

Members:

• Jacob Mobin
• Aathithya Ananth
• Benjamin DeCosta
• David Vu